The descriptions on Micro Center pages show the RX6600-6600XT as PCIe X16 spec when it is in fact X8 spec. While they do fit and operate in the X16 GPU slot they only have 8 lanes available due to the GPU hardware design by AMD. While 8 lanes may be sufficient on PCIe 4 motherboards, for now, it is not for PCIe 3 in some instances with the potential to become much worse over time. Do you feel the X16 spec in the listings is misleading and should be changed?
Micro Center updated the spec descriptions on these GPUs to accurately reflect PCIe x8 interface. Thank you for stepping up guys, well done. As the de facto DIY PC guru for my family and friends I am always emphatic about using Micro Center exclusively for all things in computing products and have been for years. I have confidently tied my reputation to yours and that was my catalyst in this post. I am so happy my confidence was not misplaced.
Good afternoon @Bodasactra
Thank you for contacting micro center. Do you have the specific graphics card you are looking at that is showing this?
It's a good suggestion. I would list it as PCIe 4.0 x16(x8 electrically) or PCIe 4.0 x8(electrically) to avoid confusion. Make it clear that it's mechanically an x16 card. We can make this suggestion to our web team.
I personally don't see the x8 wiring to being a significant bottleneck (if any) to the point where it would matter in clarifying the electrical wiring of the card itself. First and foremost, lets talk PCIe bandwidth saturation. You have PCIe 4.0 x16 which is capable of a peak bandwidth of 64GB/s. PCIe 4.0 x8 is equivalent of 32GB/s, matching PCIe 3.0 x16's bandwidth (32GB/s). There has been extensive testing by W1zzard (one of the classic GPU overclocking legends from my time back in the day) to determine if PCIe scaling matters on our current GPU's. Here are some of his results: https://www.techpowerup.com/review/nvidia-geforce-rtx-3080-pci-express-scaling/27.html
We can see that on the RTX 3080, a card physically designed to take advantage of PCIe 4.0 x16, shows only a 1% drop in performance when running at PCIe 3.0 x16 (same bandwidth as PCIe 4.0 x8). 1% is within margin of error for most testing, meaning a bottleneck does not exist in terms of PCIe bandwidth, not until we drop down to PCIe 2.0 x16 (or 3.0 x8). On the subject of PCIe 3.0 x8...
The 6600 and 6600XT being physically wired to operate at 4.0 x8 means they also must operate at 3.0 x8, because they can't take advantage of full PCIe 3.0 x16 wiring. Since we saw a 4% bottleneck on a card as fast as the RTX 3080, would it stand to reason that the 6600XT is bottlenecked as well? Not quite... Luckily W1zzard tested this as well: https://www.techpowerup.com/review/amd-radeon-rx-6600-xt-pci-express-scaling/26.html
RX 6600 XT:
We see a 2% difference in performance at 1080p, 1% performance difference at 1440p and 2% performance difference at 4K. Now W1zzard's controls are solid, so in this instance I would consider 2% to be outside margin of error, but still so small that it would be negligible for pretty much anyone not running an on-screen display and paying more attention to their framerate numbers over actual gameplay experience.
With PCIe scaling performance out of the way, lets talk about how the specs are advertised, as well as the pros and cons of changing the specifications based on this information. You will notice (at least on our website), we advertise the PCIe Interface, not the actual wiring of the card as mentioned by @TSMikeW . That is because, technically speaking, these are physically PCIe x16 interfaces on these cards, and they will not physically fit into a PCIe x8 slot unless the slot was open-ended or you took a rotary tool to your board (not recommended, but certainly doable). Most modern boards do not have PCIe x8 slots anymore, and advertising the interface as x8 will likely confuse the less-informed as to whether or not the card is compatible with their x16 slot. We have to essentially balance avoiding misinformation as well as over-complicating spec pages to avoid confusion, which can be a tricky balance to maintain.
So to answer your question of whether I believe the spec to be misleading, I would answer no, as the interface is indeed a x16 interface, despite how the card is electrically wired. I also do not believe it should be changed on the grounds of being wired x8 as we have no reason to believe this will be a significant bottleneck to warrant that kind of clarification in spec. I see it as being akin to GPU ROP, TMU, SMUs, etc, where we do not advertise those specifications despite them being the driving force in GPU performance differences in their respective classes.
With all that said, I wouldn't disagree with Mike's idea of adding an additional spec to clarify electrical wiring, I am just in the camp that it is currently unnecessary given what we know about PCIe performance scaling for now.
Submit photos and a description of your PC to our build showcase
See other custom PC builds and get some ideas for what can be done
Services starting at $149.99